1. Field of the Invention
The present invention relates to an image pickup system and image processing program which reduce random noise arising in the image pickup element system.
2. Description of the Related Art
Generally, noise components are contained in digitized signals obtained from image pickup elements and the associated analog circuits and A/D converters. Such noise components can be divided into two main categories, i.e., fixed pattern noise and random noise.
The abovementioned fixed pattern noise is noise that originates mainly in the image pickup elements, as typified by defective pixels or the like.
On the other hand, random noise is generated in the image pickup elements and analog circuits, and has characteristics that are close to white noise characteristics.
In regard to the latter random noise, for example, a technique in which the amount of noise N is converted into a function by N=abcD using constant terms a, b and c that are statically given as constant terms, and the signal level D converted into a density value, the amount of noise N for the signal level D is estimated from this function, and the filtering frequency characteristics are controlled on the basis of the estimated amount of noise N, is disclosed in Japanese Patent Application Laid-Open No. 2001-157057. Using this technique, an appropriate noise reduction treatment can be performed on the signal level.
Furthermore, as another example, Japanese Patent Application Laid-Open No. 2002-57900 discloses a technique in which the difference value A between a pixel of interest and a nearby pixel is determined, then, the mean pixel number n used in the moving average method is controlled by the function n=a/(Δ+b) using the determined difference value Δ and constant terms a and b that are statically given as constant terms, and a moving average is not determined in cases where the determined difference value Δ is equal to or greater than a specified threshold value. By using such a technique, it is possible to perform a noise reduction treatment without causing any deterioration of the original signal such as edges or the like.
However, since the amount of noise varies dynamically according to factors such as the temperature at the time of shooting, exposure time, gain and the like, conversion to a function that matches the amount of noise during shooting cannot be handled in the case of a technique using static constant terms such as that described in the abovementioned Japanese Patent Application Laid-Open No. 2001-157057, so that the precision in estimating the amount of noise is inferior. Furthermore, the filtering frequency characteristics are controlled from the amount of noise; however, since this filtering performs equal processing without discriminating between flat portions and edge portions, the edge portions in regions, where it is estimated on the basis of the signal level that the amount of noise is large, deteriorate. Specifically, processing that discriminates between the original signal and noise cannot be handled, so that the preservation of the original signal is poor.
Furthermore, in the technique described in Japanese Patent Application Laid-Open No. 2002-57900, the determination of whether or not the moving average method is performed is accomplished by comparison with a threshold value. However, since this threshold value is also given statically, variation in the amount of noise according to the signal level cannot be handled, so that the selection of the average number of pixels or moving average method cannot be optimally controlled. Consequently, noise components remain, resulting in deterioration of the original signal and the like.